Required Roles of Bax and JNKs in Central and Peripheral Nervous System Death of Retinoblastoma-deficient Mice*

Retinoblastoma-deficient mice show massive neuronal damage and deficits in both CNS and PNS tissue. Previous work in the field has shown that death is regulated through distinct processes where CNS tissue undergoes death regulated by the tumor suppressor p53 and the apoptosome component, APAF1. Death in the PNS, however, is independent of p53 and reliant on the death protease, caspase 3. In the present study, we more carefully delineated the common and distinct mechanisms of death regulation by examining the stress-activated kinases, JNK2 and 3, the conserved Bcl-2 member Bax, and the relationship among these elements including p53. By use of genetic modeling, we show that death in various regions of the CNS and DRGs of the PNS is reliant on Bax. In the CNS, Bax acts downstream of p53. The relevance of the JNKs is more complex, however. Surprisingly, JNK3 deficiency by itself does not inhibit c-Jun phosphorylation and instead, aggravates death in both CNS and PNS tissue. However, JNK2/3 double deficiency blocks death due to Rb loss in both the PNS and CNS. Importantly, the relationships between JNKs, p53, and Bax exhibit regional differences. In the medulla region of the hindbrain in the CNS, JNK2/3 deficiency blocks p53 activation. Moreover, Bax deficiency does not affect c-Jun phosphorylation. This indicates that a JNK-p53-Bax pathway is central in the hindbrain. However, in the diencephalon regions of the forebrain (thalamus), Bax deficiency blocks c-Jun activation, indicating that a Bax-JNK pathway of death is more relevant. In the DRGs of the PNS, a third pathway is present. In this case, a JNK-Bax pathway, independent of p53, regulates damage. Accordingly, our results show that a death regulator Bax is common to death in both PNS and CNS tissue. However, it is regulated by or itself regulates different effectors including the JNKs and p53 depending upon the specific region of the nervous system.

The tumor suppressor Retinoblastoma (Rb) 2 influences a wide number of biological processes including cell growth, modulation of cell death, and differentiation (1). Rb is mutated in a third of human tumors and its role in oncogenesis is intensively studied (2). Its importance is further underscored by observations that Rb mutant mice display gross neuronal defects accompanied by ectopic S phase entry, and midgestation (E13-E14) lethality (3)(4)(5). Cell death is observed in Rb mutant embryos in PNS, CNS, muscle, lens, and fetal liver tissues.
Recent efforts have focused on understanding the mechanisms of neuronal loss induced in both PNS and CNS of developing Rb mutant mice. Neuronal death in these mice appears to occur through a complex tissue-specific signaling cascade, making this an intriguing model of death that occurs during development. Death is not cell autonomously regulated but likely due to other secondary events such as hypoxia (6,7). In this paradigm, neuronal death in the PNS and CNS also appear to be regulated differentially. Death in the CNS is regulated by E2F1 and E2F3, two Rb-regulated transcription factors critical for cell cycle regulation (8 -10). In addition, death in the CNS is entirely dependent upon the tumor suppressor p53 (11), and Apaf1 (12), a central component of the apoptosome critical for activation of the mitochondrial pathway of apoptosis. Finally, death is not affected by caspase 3 deficiency (13).
In contrast to the CNS, neuronal loss in PNS tissue of Rbdeficient mice is only partially dependent upon E2F1/3 (8 -10) and Apaf1 (12), but completely dependent upon caspase 3 (13) and independent of p53 (11). These data suggest that both proximal (e.g. p53) and distal (e.g. caspase 3) effectors of death differ in the CNS and PNS. These findings highlight the important question of the identity(ies) of the central common mechanistic elements between CNS and PNS death in Rb-deficient embryos. As will be discussed in more detail below, Bax is thought to be critical in many forms of neuronal death. Accordingly, we initially hypothesized that the Bcl-2 member, Bax might be the central death regulator. We also postulated that there were different upstream regulators of Bax in the CNS and PNS tissue. In this regard, we examined both p53 and the JNKs, two known upstream regulators of Bcl-2 family members.
The Bcl-2 family member Bax has been shown to be central in neuronal death in a number of different paradigms (for example (14 -16)). In many cases, Bax translocation to the mitochondria leads to release of cytochrome c, concomitant activation of the apoptosome, and in turn, activation of numerous effector caspases. However, Bax does not necessarily mediate death solely through activation of apoptosome. Bax may also regulate apoptosis-inducing factor (AIF) release from the mitochondria (17). As such, Bax represents a viable candidate to explain the Apaf1-independent death processes observed in the PNS. The factors which lead to activation of Bax are numerous. Of relevance, p53 (14) as well as the stress activated kinases JNKs (18 -22) have been shown to regulate Bax and/or other death promoting Bcl-2 family members in neurons. Therefore, Bax could play a central role in mediating both p53/Apaf1-dependent death (as in the CNS) and p53/Apaf1-independent death as might occur in the PNS.
JNKs, members of the larger MAPK superfamily, are a family of three (JNK1,2,3) kinases which are activated by multiple stimuli including death inducing insults such as growth factor deprivation, DNA damage, dopaminergic toxins, axotomy, and ischemic insult (18,(23)(24)(25)(26)(27)(28)(29)(30). The JNK family members are key activators of the transcription factor c-Jun by phosphorylating c-Jun on Ser 63 and Ser 73 (31)(32)(33). In most cases in neurons, JNK2 and 3 have been primarily associated with death (29,30). While the JNK/c-Jun pathway has been shown to regulate activation of Bcl-2 family members such as Bim in models of death that are Bax-dependent, JNKs can also modulate p53 function (37,38). The interplay between p53, JNK and Bax is therefore likely complex. We presently set out to delineate the potentially critical role of these signals in neuronal death due to Rb deficiency.
By use of genetic modeling using Bax-and JNK-deficient mice, we report that death in dorsal root ganglia (DRGs) of the PNS and various regions of the CNS is Bax-dependent. However, the relationship of JNKs to death is more complex. JNK3 deficiency alone exacerbates death in PNS and CNS tissue. However, JNK2/3 double deficiency is protective. Our data also provide a model by which JNKs act differentially in the CNS. JNKs can regulate both p53 and c-Jun upstream of Bax in the caudal region of the hindbrain (medulla oblongata). However, it can also act downstream of Bax in the diencephalic regions of the forebrain. In the DRGs of the PNS, death is dependent upon JNKs which acts upstream of Bax, independent of p53. Therefore we propose that a common central activator of death, Bax, is regulated by and activates differential upstream and effector signaling pathways in the CNS and PNS tissues of Rb-deficient mice.
For TUNEL and immunofluorescence analyses, embryos were obtained at E13.5. Tissue samples were obtained for genotyping purposes. The embryos were fixed in 4% PFA/0.1 M phosphate buffer pH 7.4 overnight at 4°C. The embryos were rinsed using 1ϫ PBS, pH 7.4 (Invitrogen) and cryoprotected in 30% sucrose/0.1 M phosphate buffer pH 7.4 overnight at 4°C. Embryos were equilibrated in a 50:50 mixture of 30% sucrose/ OCT (TissueTek 4583) for ϳ1 h on a rocking platform. Finally, the embryos were embedded in the same 50:50 mixture and frozen on liquid nitrogen. The embryos were sectioned sagitally at 14 m on a cryostat. In the Bax/Rb breedings, a subset of embryos was also collected at E18.5 to determine whether survival was enhanced with Bax deficiency.
TUNEL Labeling-TUNEL labeling of sections was performed as previously described (13). Briefly, sections were fixed with 1% glutaraldehyde for 15 min. Following washing with PBS, sections were permeabilized with 1:1 methanol/acetone mixture for 10 min. After washing with PBS, slides were incubated with Hoechst (1:4000 of 0.5 mg/ml soln), and TUNEL reaction was performed for 1 h 37°C using the terminal transferase kit as per the manufacturer's instructions (Roche Applied Science). TUNEL labeling was visualized by secondary labeling using Cy3-streptavidin (Jackson Laboratories). For quantification, regions of the thalamus in the forebrain, medulla in the hindbrain, trigeminal ganglion and caudal DRGs were quantitated as follows: areas were photographed using Northern Eclipse, Empix Imaging software. Magnification: 20 ϫ 4 ϫ 5 grid set up, positive neurons were counted and compared with their respective Hoechst-positive nuclei. Both TUNEL labeling as well as the total number of Hoechst-positive cells were evaluated. The data are expressed as TUNEL-positive cells/total Hoechst positive cells Ϯ S.E. Statistical significance was analyzed by analysis of variance with Newman-Keuls multiple comparison.
Immunofluorescence Analysis-Sections were obtained as described above and immunofluorescence described previously (13). Briefly, sections were washed three times in PBS, incubated overnight in PBS containing 0.3% Triton X-100 and primary antibody. Primary antibodies utilized were c-Jun (Cell Signaling Technologies, 1:150), phospho-c-Jun Ser 73 (Cell Signaling 1:150), and phospho-Ser 15 p53 (Cell Signaling, 1:150), Nestin (RDI, 1:400), Active caspase 3 (BD Pharmingen. 1:500), B-III Tubulin (kind gift from D. Brown, 1:50). Following three washes in PBS, sections were incubated in PBS, 0.3% Triton X-100 solution containing Cy3-conjugated donkey anti rabbit secondary antibody (Jackson, 1:2000). Quantitation was performed similar to that described above for TUNEL analysis. In the case of active caspase 3, total counts were evaluated in relation to a defined region and not to the number of Hoechst cells to facilitate analyses.

Bax Deficiency Rescues PNS and CNS Death in Rb-deficient
Mice-To determine whether Bax acts as a common mediator of PNS and CNS neuronal death in Rb-deficient neurons, we interbred Bax heterozygous deficient mice with RB heterozygous deficient mice to obtain double heterozygous progeny. These animals were then bred to produce double homozygous-deficient mice. Alternatively, Bax/Rb double heterozygous mice were also interbred with Bax-deficient Rb heterozygous mice. The expected and actual frequencies of genotypes from E13.5 embryos obtained from these crosses are listed in Table 1. The frequencies of embryos were roughly as expected. The majority of the analyses were performed at this age since gross apoptosis in both PNS and CNS is observed at this point prior to embryonic lethality of Rb-deficient mice.
As shown in Supplemental Fig. S1, widespread death was observed in a variety of areas of the Rb-deficient CNS as previously reported (3)(4)(5). Death in the cortex (dorsal telencepha-lon) ganglionic eminences (ventral telencephalon), thalamus (diencephalon), and pons and medulla of the hindbrain was evident as visualized by TUNEL (data not shown) or staining with active caspase 3 antibody. Active caspase 3 was also observed in these regions in both B-III tubulin-positive (early postmitotic neurons) and -negative regions as shown by double labeling analyses (Supplemental Fig. S1). Active caspase 3 positivity was also widespread in nestin positive (progenitor and early postmitotic) cells (Supplemental Fig. S2). We also examined a select population of neurons using calbinden antibody which detects ventral telencephalon-derived interneurons (Supplemental Fig. S3). In this case, we could not colabel with anticaspase 3 (same antibody species as calbinden) but we instead colocalized with a condensed fragmented hoechst phenotype indicative of damage. Colocalization was observed in a number of regions as stated above. Our results show that widespread cell death occurs in the progenitor (B-III-tubulin negative) and early postmitotic (B-III-tubulin positive) regions of the developing dorsal (cortex) and ventral (lateral and medial ganglionic eminences) telencephalon, diencephalons (thalamus), roof of the midbrain and hindbrain (pons and medulla). In the PNS, death was observed in all DRG ganglia. However, less significant death was observed in superior cervical ganglia of the sympathetic system, suggesting that death in the PNS due to Rb deficiency is more prominent in sensory ganglia (Supplemental Fig. S4).
To examine the effect of Bax deficiency on death due to Rb deletion, TUNEL analysis was performed on whole embryo sections of the genotypes described above at E13.5. Because death is widespread, only select regions of the CNS (thalamus of the forebrain, medulla of the hindbrain) were quantified. In addition, because no significant death was observed in SCGs, only sensory ganglia were examined. In this regard, we quantified death in the trigeminal ganglia and caudal DRGs. As shown in Fig. 1, control (double heterozygous, WT heterozygous, or double WT) embryos displayed occasional TUNEL labeling in both PNS and CNS tissues examined. Rb deficiency resulted in massive apoptosis in both tissues examined when compared with control animals. For example, 8% death (TUNEL positivity) was observed in the forebrain (thalamus) and 8.3% in the hindbrain (medulla) regions of the CNS of Rb-deficient mice as compared with 0 and 0.8% death in control animals, respectively ( Fig. 2A). Death was also observed by Hoechst analyses where there was significant overlap between TUNEL-positive neurons and condensed nuclei. In the PNS, increased death in the trigeminal ganglion (21 versus 2% control) and rostral DRG nuclei (14 versus 5% in controls) was also evident in Rb-deficient embryos ( Fig. 2A). TUNEL labeling in Bax/Rb double-deficient embryos, however, were dramatically reduced in all neuronal populations examined. In the thalamus and medulla regions of the CNS, death was reduced (8.4% Rb-deficient versus 3.3% double-deficient forebrain; (8.3% RB-deficient versus 2.3% double-deficient hindbrain). Similar reductions were also observed in the PNS (20.8 versus 5.3% in trigeminal and 13.8 versus 4.8% in rostral DRGs). It is important to note that Bax deficiency appeared to affect death in both B-III tubulin positive early neuronal and negative progenitor regions. For example, in the dorsal cortex and midbrain regions where the distinction between B-III tubulin-positive and -negative areas are the most defined, we observed that active caspase 3 was reduced in both tubulin-positive and -negative regions ( Fig. 2B) with Bax/Rb double-deficient embryos. This was not the case with JNK3 deficiency, which is not protective against death due to Rb loss as described below. Because Bax deficiency dramatically reduced neuronal death in the brain regions examined, we asked whether Bax deletion would promote increased survival of Rb-deficient embryos. Accordingly, the number of Bax/Rb double-deficient embryos was analyzed at E18.5. However no viable Rb-deficient animals were found at this age (n ϭ 39).
Previous evidence has indicated that p53 is induced in the CNS of Rb-deficient mice (11). In addition, cell death in this region is dependent upon p53 since p53/Rb double-deficient mice show less death when compared with Rb-deficient mice alone (11). Therefore, we next determined whether Bax may inhibit p53 activation or whether protection mediated by Bax would leave the p53 response intact. The latter would be expected if p53 acted upstream of Bax. As shown in Fig. 3, active phosphorylated p53 was detected utilizing a phosphospecific phosphoSer 15 p53 antibody. Modification at this site has been associated with increased stabilization/activation of p53. Phosphorylated p53 is most clearly induced in the medulla of the CNS in Rb-deficient but not control mice. No significant induction of active p53 was observed in the DRGs and only minor induction in the thalamus (data not shown). In contrast to the protection observed with Bax deficiency, however, Bax/Rb-deficient mice showed no significant decrease in p53 levels in the hindbrain (medulla). Similar results were also obtained for total p53 staining in the medulla (Fig. 3D). Finally, it was also important to show that activated p53 also occurred in dying cells in Rb-deficient mice and that this was reversed with Bax deficiency. To do this, we quantified the percentage of phospho-p53-positive cells, which also showed signs of nuclear condensation and fragmentation in the medulla. Our results show that ϳ30% was colocalized (Fig. 3F). Colabeling is not 100% because p53 induction likely precedes late death markers such as nuclear condensation. We then determined whether this colocalization was altered in Rb/Bax double-deficient animals. In fact, we observed that the colocalization with markers of damage decreased dramatically (less than 10%) as a result of Bax   deficiency. This reduction did not occur with JNK3 deficiency which is not protective against cell death (see below for full results on JNKs). This is strong supportive evidence that cells which are p53 positive are the ones which are destined to die. Taken together, our data suggest that at least in the medulla, Bax acts at a point downstream of p53 and that inhibition of Bax still leaves p53 activation intact in this region.
The Role of JNKs in RB-mediated Death-The JNK/c-Jun pathway is reported to affect neuron survival under select conditions. In peripheral neurons such as sympathetic neurons deprived of NGF, it is thought that this pathway regulates Bax activation (18 -21). Because Bax is a common mediator of death in both PNS and CNS tissues evaluated, we examined whether the JNK/c-Jun pathway might regulate neuronal death in both regions. We first examined whether c-Jun is phosphorylated on Ser 73 , a known JNK phosphorylation site in Rb-deficient mice in the selected PNS and CNS regions as described above. As shown in Fig. 4, a substantial increase in phosphorylated c-Jun was observed in both forebrain and hindbrain of the CNS examined as well as in TG and rostral DRGs of the PNS. This induction of c-Jun phosphorylation observed in Rb-deficient mice was not affected in the sensory ganglia by Bax deficiency. This suggests that in this tissue, JNKs act upstream of Bax. In the CNS, however, Bax deficiency blocks c-Jun phosphorylation observed in Rb-deficient mice in the thalamus, but not in the medulla. This indicates that in various regions of the CNS, the linkage between Bax and JNKs differ. In the medulla, JNKs, like p53 likely act more proximally prior to Bax activation. However in the thalamus, Bax appears to regulate c-Jun phosphorylation. Similar results were obtained for total c-Jun staining (Fig. 4). Importantly, similar to the results discussed above for p53, c-Jun phosphorylation colabeled to a significant degree with markers of nuclear condensation/fragmentation in Rb-deficient cells (Fig. 4D). As shown in Fig. 4D, this colocalization was dramatically reduced with Bax deficiency, which is protective, but not with JNK3 deficiency, which fails to protect (see below). This supports the notion that c-Jun phosphorylation occurs in dying cells.
We next asked whether JNKs might be important for neuronal death observed in Rb-deficient embryos. To do this, we first examined the effects of loss of JNK3, the JNK isoform present predominately in the brain and most associated with neuronal death (29,30). We interbred JNK3 heterozygous-deficient mice with Rb heterozygous-deficient mice to obtain double heterozygous progeny. These animals were then bred to produce double homozygous-deficient mice. Alternatively, JNK3/Rb double heterozygous mice were also interbred with JNK3-deficient Rb heterozygous mice. The expected and actual frequencies of genotypes from E13.5 embryos obtained from these crosses in combination are listed in Table 2. As shown for Bax/Rb embryos (Table 1), the frequencies of embryos for all JNK3 and Rb genotype combinations were roughly as expected.
We next determined whether JNK3 deficiency protects from neuron damage induced by Rb loss in the DRGs of the PNS and select CNS regions by TUNEL analysis. In this regard, JNK3/RB double-deficient embryos did not show reduced TUNEL reactivity when compared with Rb-deficient embryos alone in any CNS or PNS regions (Figs. 1 and 5). In fact, JNK3 deficiency caused a dramatic increase in death caused by Rb deficiency in both CNS (hindbrain-medulla 19% forebrain-thalamus 17 versus 6.5 and 9.5%, respectively) and PNS (DRG 41 versus 14%) tissues (Fig. 5). This suggests that JNK3 by itself may even have some protective function or that there is compensation by other JNK members in the absence of Rb. No other gross phenotype other than increased TUNEL was detected in Bax/Rb double knock-out embryos when compared with just Rb-deficient animals. JNK3 deficiency by itself had little effect in death in CNS and PNS tissues. Similarly, JNK2 deficiency also failed to provide significant protection in the PNS, but did provide a slight protective effect in the CNS (Fig. 5). Importantly, when compared with JNK3 deficiency, JNK2 deficiency did not result in increased TUNEL positivity.
Because JNK 2 and 3 family members may compensate for each other, we also determined whether a deficiency of both JNK2 and 3 would ameliorate death induced by loss of Rb. As shown in Fig. 5, death in the sensory PNS was reduced in JNK2/ 3/Rb triple-deficient mice when compared with Rb deficiency alone (14% versus 6%). Similarly, death in the CNS regions examined was also significantly reduced in JNK2/3/Rb-deficient embryos. This decrease was more dramatic when com-

TABLE 2 Genotype frequency of offspring from all JNK3 Rb interbreedings
Two type of breedings were performed. These include double heterozygous crosses as well as breedings of double heterozygous mice with JNK Ϫ/Ϫ RB ϩ/Ϫ mice. Each of these crosses give different expected Mendelian ratios. The expected calculations listed here are ratios of these strategies combined taking into consideration the number of breedings for each strategy. pared with JNK2/Rb double-deficient embryos in the CNS. This indicates that death in the PNS and CNS is dependent upon the combination of JNK2/3. We next determined how JNK deficiency is associated with the level of c-Jun and c-Jun phosphorylation. As shown in Fig. 6, JNK3 deficiency, which did not prevent the loss of neurons in Rb-deficient mice, also did not prevent c-Jun phosphorylation or c-Jun increase in any tissues examined. Interestingly, in some tissues (sensory DRG, Forebrain-thalamus), JNK3 deficiency by itself appeared to increase c-Jun levels. The reason for this is unclear, but could be one explanation of why JNK3 deficiency might exacerbate damage as shown in Fig. 5. JNK2 deficiency alone or JNK2/3 double deficiency significantly blocked c-Jun phosphorylation or c-Jun increase in the CNS and PNS tissue of Rb-deficient mice. Interestingly, JNK2 deficiency by itself fails to block death in the PNS while at the same time blocking c-Jun phosphorylation. This indicates that at least in this tissue, factors other than c-Jun phosphorylation are critical for death.

Genotype of offspring
The JNKs are also thought to mediate activation of p53 under select conditions (37,38). Therefore we asked whether p53 may act downstream of JNKs in the hindbrain, the region where p53 activation is most pronounced. Similar to TUNEL labeling, JNK3 deficiency failed to down-regulate p53 (Fig. 3,  B and E.). However, JNK 2-or JNK2/3-deficient mice did show reduced phospho-p53 and total p53 labeling (Fig. 3, B  and E). This suggests that JNK2/3 act upstream of p53 activation in the hindbrain.

DISCUSSION
Rb-deficient embryos die early around E13.5 and display massive neuronal loss in various regions of the CNS and PNS. As such, this model is an important system to study death processes, which occur in the developing nervous system (3)(4)(5). The mechanism(s) by which neuronal death occurs in this paradigm is not fully defined but there is increasing evidence that the pathways differ in the two tissues. Accordingly, we embarked to determine whether there might be common elements in the death pathways of the CNS and PNS and whether they may be additional death promoting factors, which might explain the different regulatory pathways. The present results indicate a number of conclusions as follows: 1) Bax is a common critical factor regulating death in both DRGs of the PNS and CNS tissues examined. 2) However, Bax does not regulate all cell loss due to Rb deficiency and does not extend the life of the embryo. 3) Bax deficiency does not affect p53 activation suggesting that it acts downstream of p53 in the medulla of the CNS. 4) Interestingly, Bax deficiency also promotes a reduction in c-Jun activation, but only in the thalamus (in the forebrain) of the CNS. 4) JNK3 deficiency exacerbates death in both the CNS or PNS tissues examined. However, double-deficient JNK2/3 mice display reduced death in these tissues. 5) This protection is correlated with reduced c-Jun phosphorylation in the CNS areas examined, but not the sensory ganglia of the PNS. 6) JNK2/3 deficiency also reduces p53 activation in the medulla (hindbrain) of the CNS. Taken together our results suggest a region specific linkage of the JNKs, Bax, p53, and c-Jun. We propose a model by which death in the DRGs of the PNS is regulated by a JNK-Bax pathway, which acts to promote death in an Apaf/caspase 3-dependent fashion. This pathway is also partially dependent upon the E2Fs, although the linkage at this point is not clear. Death in the CNS is region specific. In the medulla-hindbrain, JNKs regulate both p53 and c-Jun activation which acts upstream of Bax. In this region, E2Fs also regulate p53. Death is promoted more distally by an Apaf1 dependent pathway independent of caspase3. In the thalamus-forebrain, in contrast, JNK/c-Jun involvement appears downstream of Bax (see Fig. 7).
CNS Pathway-Death in the CNS by Rb deficiency has been previously reported to be dependent upon both E2F1 and E2F3 (8 -10). Mice deficient for both Rb and E2Fs show almost complete inhibition of death and ectopic proliferation observed in the CNS (8 -10). This observation is consistent with the well established role of E2Fs as Rb targets (39). However, it is important to note that death in the nervous system due to Rb deficiency is not cell autonomous and likely due to secondary events such as hypoxia (6,7). Therefore, it is unclear whether E2Fs mediate death in the CNS directly in damaged neurons or whether E2F deficiency rescues more primary defects associated with RB loss. In the CNS, cell death of Rb-deficient embryos is also completely dependent upon p53 (11). This p53- dependent death is also associated with a dependence on the ced4 homologue Apaf1 (12).
Presently, we show that Bax is also a critical mediator of CNS death in the Rb-deficient embryo. Rb/Bax-deficient embryos show dramatically reduced TUNEL labeling when compared with Rb-deficient controls in both hindbrain and forebrain regions examined. Bax deficiency however, does not affect p53 induction suggesting that Bax acts downstream of p53. These results are consistent with our previous reports indicating that Bax is required for p53-mediated death in several neuronal death paradigms in vitro and that Bax can also regulate activation of the apoptosome (14,16).
The JNKs also appear critical to CNS death. However, the mechanism by which it does so is more complex and region specific. First, the complexity is exacerbated by the fact that there are multiple JNK members. We focused on JNK2 and 3 since they appeared to be most associated with neuronal loss. Surprisingly, depletion of JNK3 alone not only failed to protect, but also exacerbated Rb deficiency mediated damage. This is consistent with the point that JNK is not universally associated with neuronal death. Several reports have indicated that the JNK/c-Jun pathway can also be important for axonal regeneration and process outgrowth under select conditions (34 -36). The dual nature of JNK participation in death/survival is strengthened by our present JNK3 deficiency data. However, whether our observations may be due to a compensation by other JNK family members or loss of an important survival function of the JNK3 isoform is currently unknown. Depletion of JNK2 and 3, however, leads to significant reduction in death in both hindbrain and forebrain regions of the CNS examined. These results are consistent with previous reports suggesting the importance of JNK isoforms in neuronal death. For example, JNK isoforms have been associated with death in models of stroke and ischemia (29,30). A final level of complexity is illustrated by how JNKs may be related to Bax activation in the CNS. In the forebrain, Bax deficiency abolishes c-Jun activation indicating that the JNK/c-Jun pathway is downstream of Bax. However, in the hindbrain, this is not the case and the linkage is reversed.
Interestingly, while death appears to be regulated by the apoptosome in the CNS, caspase 3 deficiency by itself is not sufficient to inhibit CNS damage (13). This suggests that Baxmediated death signal likely activates and utilizes multiple caspases in the CNS to cause damage. From our present data as well as of others, we propose a model by which E2Fs, either in a direct cell autonomous or indirect fashion leads, to activation of Bax. However, how Bax is regulated and the consequence of Bax regulation depends upon on the region of the CNS. A caveat to our studies as well as those of others is that we have not firmly established the presence/absence of differences, which might occur between cell types in the areas of the CNS. However, we and others have observed that death is widespread between progenitors and newly differentiated neurons and have not noticed any qualitative changes in these two populations.
PNS Pathway-In contrast to the CNS, death in the DRGs of the PNS is less dependent upon the E2Fs and is independent of p53 (8 -11). Mice which are double-deficient for both Rb and p53 do not display a reduction in damage when compared with Rb-deficient embryos alone (11). Interestingly, death in sensory ganglia is only partially dependent upon Apaf1 suggesting that death can also occur in independent of the classic apoptosome (12). As in the CNS, we presently show that Bax deletion abolishes death, which results from Rb deficiency in the PNS. This suggests that Bax plays a more central role than Apaf1. We propose that Bax in fact may mediate Apaf1-dependent and -independent processes. In support of this, Bax has been shown to participate in alternative apoptosome independent modes of neuronal death in vitro. These pathways include regulation of alternative mitochondrial factors, which also promote death such as AIF (17).
Death in the CNS appears to proceed through a JNK-p53-Bax-Apaf1-or Bax-JNK/APAF1-dependent process. In contrast, the proximal signals which regulate the mitochodrial pathway of death in the sensory ganglia of the PNS is independent of p53 but instead rely only upon the JNKs. Depletion of a combination of JNK2 and 3 reduces TUNEL labeling in the PNS associated with RB deficiency. The observation that Bax deficiency does not affect c-Jun phosphorylation also suggests that JNK acts upstream of Bax activation. Interestingly, the functional relevance of c-Jun phosphorylation by the JNKs in medi-ating PNS death is questionable. For example, our data show that JNK2 deficiency appears to inhibit c-Jun phosphorylation in the PNS, but does not prevent death. This is not the case in the CNS where inhibition of c-Jun and protection by JNK deficiency is tightly correlated. Again, this suggests that c-Jun has multiple potential functions outside of promotion of cell death (34 -36, 40).
In summary, our data, in conjunction with other reports, support a model by which death in the PNS is initiated by a JNK/Bax pathway. However, death in the CNS is more complex where the involvement of JNK and p53 in Bax-mediated death differs depending upon the region of the brain. Once activated Bax likely regulates multiple pathways including but not limited to that of Apaf1/caspase 3. These results not only suggest common elements in neuronal loss in CNS and PNS tissue, it also delineates how these death elements differentially control damage in distinct regions of the nervous system.